MDP: Markov decision making process (III.)

MDP: Markov decision-making processes (Markov decision process)

Behrman equation:

In the previous section, this is a model of deterministic action. If the model of the random action, it should be expressed as

That is, there are multiple states after performing an action, and the probability is multiplied by the value function to get the formula.

Therefore, the current state optimal execution action is

There is a vπ (S) for each state, so for each step, you can get the following recursive form:

MDP: Markov decision making process (III.)