-A decision maker observes a discrete-time system which moves between states {S1, S2, S3, S4} according...

 

Transcribed Image Text:

-A decision maker observes a discrete-time system which moves between states {S1, S2, S3, S4} according to the following transition probability matrix: 0.3 0.2 0.1 0.4 P = 0.4 0.2 0.1 0.5 0.0 0.3 0.0 0.8 0.1 0.0 0.0 0.6 At each point in time, the decision maker may leave the system and receive a reward of R = 20 units, or alternatively remain in the system and receive a reward of r(s;) units if the system occupies state s;. If the decision maker decides to remain in the system its state at the next decision epoch is determined by matrix P. Assume a discount rate of 0.9 and that r(s;) = i. a) Formulate this model as an MDP. b) Use both policy iteration and linear programming to find a stationary policy which minimizes the expected total discounted reward. compare the results, and report the optimal policy and the optimal value function for both methods. c) Find the smallest value of R so that it is optimal to leave the system in state 2. -A decision maker observes a discrete-time system which moves between states {S1, S2, S3, S4} according to the following transition probability matrix: 0.3 0.2 0.1 0.4 P = 0.4 0.2 0.1 0.5 0.0 0.3 0.0 0.8 0.1 0.0 0.0 0.6 At each point in time, the decision maker may leave the system and receive a reward of R = 20 units, or alternatively remain in the system and receive a reward of r(s;) units if the system occupies state s;. If the decision maker decides to remain in the system its state at the next decision epoch is determined by matrix P. Assume a discount rate of 0.9 and that r(s;) = i. a) Formulate this model as an MDP. b) Use both policy iteration and linear programming to find a stationary policy which minimizes the expected total discounted reward. compare the results, and report the optimal policy and the optimal value function for both methods. c) Find the smallest value of R so that it is optimal to leave the system in state 2.