In reinforcement learning, the ecosystem is often represented as a Markov decision process (MDP). Quite a few reinforcements learning algorithms use dynamic programming techniques.[fifty three] Reinforcement learning algorithms tend not to presume knowledge of a precise mathematical model of the MDP