Interface PolicyEvaluation<S extends State, A extends Action>

Type Parameters:
S - The type of states in the MDP.
A - The type of actions in the MDP.
All Known Implementing Classes:
IterativePolicyEvaluation
public interface PolicyEvaluation<S extends State, A extends Action>
The `PolicyEvaluation` interface provides methods to evaluate the utility of states in a Markov Decision Process (MDP) with respect to a given policy.
Author:
Matthias Thimm

  • Method Details

    • getUtilities

      Map<S,Double> getUtilities(MarkovDecisionProcess<S,A> mdp, Policy<S,A> pi, double gamma)
      Determines the utilities of the states in the MDP wrt. the given policy.
      Parameters:
      mdp - some MDP
      pi - some policy
      gamma - the discount factor
      Returns:
      the utilities of the states of the MDP.