Markov Decision Process (MDP) – Ch5 10 B F 6. Consider the MDP above, with states represented as nodes and transitions as edges between nodes. The rewards for the transitions are indicated by the...

Extracted text: Markov Decision Process (MDP) – Ch5 10 B F 6. Consider the MDP above, with states represented as nodes and transitions as edges between nodes. The rewards for the transitions are indicated by the numbers on the edges. For example, going from state B to state A gives a reward of 10, but going from state A to itself gives a reward of 0. Some transitions are not allowed, such as from state A to state B. Transitions are deterministic (if can choose to go from one to the other and will there is an edge between two states, the reach the other state with probability 1). A. Suppose that the max horizon length is 15, write down the optimal action at each step if the discount factor is y = 1.


Extracted text: B. Now suppose that the horizon is infinite. For each state, does the optimal action depend on y? If so, for each state, write an equation that would let you determine the value for y at which the optimal action changes.