Type Members

1.  trait ActionValueFn[Obs, A, T] extends AnyRef

   Along with StateValueFn, this is the main trait in tabular reinforcement learning for tracking the value of an (observation, action) pair.

   Along with StateValueFn, this is the main trait in tabular reinforcement learning for tracking the value of an (observation, action) pair.

   Obs

   Observation returned by the State instances tracked by ActionValueFn.

   A

   Action type on the State instances tracked by ActionValueFn.

   T

   type of values tracked by ActionValueFn.

2.  trait Agent[Obs, A, R, T, M[_]] extends AnyRef
3.  type Cat[+T] = Categorical[T]
4.  trait Policy[Obs, A, R, M[_], S[_]] extends AnyRef

   This is how agents actually choose what comes next.

   This is how agents actually choose what comes next. This is a stochastic policy. We have to to be able to match this up with a state that has the same monadic return type, but for now it's hardcoded.

   A - Action Obs - the observation offered by this state. R - reward M - the monadic type offered by the policy. S - the monad for the state.

5.  final case class SARS[Obs, A, R, S[_]](state: State[Obs, A, R, S], action: A, reward: R, nextState: State[Obs, A, R, S]) extends Product with Serializable

   Chunk that you get back for playing an episode.

6.  trait State[Obs, A, R, M[_]] extends AnyRef

   A world should probably have a generator of states and actions...

   A world should probably have a generator of states and actions... and then you can use that to get to the next thing. The state here is going to be useful in the Markov model; for the bandit we only have a single state, not that useful.

7.  trait StateValueFn[Obs, T] extends AnyRef

   Along with ActionValueFn, this is the main trait in tabular reinforcement learning for tracking the value of a state as evidenced by the observation it returns.

   Along with ActionValueFn, this is the main trait in tabular reinforcement learning for tracking the value of a state as evidenced by the observation it returns.

   We need some way for this to learn, or see new observations, that's part of the trait.

   Obs

   Observation returned by the State instances tracked by StateValueFn.

   T

   type of values tracked by StateValueFn.

8.  final case class Time(value: Long) extends AnyVal with Product with Serializable