The reinforcement learning environments defines the world where the RL-agent(decision maker) gain experience by taking action that changes the environment, this action could induce a positive change environment or even become adverse.
The changes in the system(environment) is provided to RL-agent via. `scalar` quantity, and the RL-agents updates itself based on this `reward` signal.
graph LR
A[Agent] -- action --> B((Environemt))
B -- reward, state --> A