Description
This project has code to compute an optimal policy for a given MDP using the algorithms that were discussed in the course: Value Iteration, Howard’s Policy Iteration, and Linear Programming (using PuLP). Input to these algorithms will be an MDP and the expected output is the optimal value function, along with an optimal policy. There is an optional command line argument for the policy, which evaluates the value function for a given policy instead of finding the optimal policy, and returns the action and value function for each state in the same format.
MDP solvers have a variety of applications. The second part of this project uses the solver to find an optimal policy for a batter chasing a target during the last wicket in a game of cricket.