Package burlap.behavior.policy

Interface Summary

Interface	Description
EnumerablePolicy	An interface extension to Policy for policies that can enumerate their probability distribution over all actions.
Policy	An interface for defining a Policy.
SolverDerivedPolicy	An interface for defining policies that refer to a MDPSolverInterface objects to defined the policy.

Class Summary

Class	Description
BoltzmannQPolicy	This class implements a Boltzmann policy where the the Q-values represent the components of the Boltzmann distribution.
CachedPolicy	This class can be used to lazily cache the policy of a source policy.
EpsilonGreedy	This class defines a an epsilon-greedy policy over Q-values and requires a QComputable valueFunction to be specified.
GreedyDeterministicQPolicy	A greedy policy that breaks ties by choosing the first action with the maximum value.
GreedyQPolicy	A greedy policy that breaks ties by randomly choosing an action amongst the tied actions.
PolicyUtils
RandomPolicy	A uniform random policy for single agent domains.