reinforce-0.0.0.1: Reinforcement learning in Haskell

Copyright(c) Sentenai 2017
LicenseBSD3
Maintainersam@sentenai.com
Stabilityexperimental
Safe HaskellNone
LanguageHaskell2010

Environments.Gym.ToyText.FrozenLakeV0

Description

The agent controls the movement of a character in a grid world. Some tiles of the grid are walkable, and others lead to the agent falling into the water. Additionally, the movement direction of the agent is uncertain and only partially depends on the chosen direction. The agent is rewarded for finding a walkable path to a goal tile.

https://gym.openai.com/envs/FrozenLake-v0

Synopsis

Documentation

type Runner s a x = RunnerT s a IO x Source #

type alias of RunnerT in IO

newtype StateFL Source #

The current position of the agent on the frozen lake

Constructors

Position 

Fields

Instances

toVector :: StateFL -> Vector Int Source #

Convert StateFL to a computable type

mkStateFL :: MonadThrow m => Int -> m StateFL Source #

Build a FrozenLakeV0 state, throwing if the position is out of bounds.

type EnvironmentT t = GymEnvironmentT StateFL Action t Source #

Alias to GymEnvironmentT with FrozenLakeV0 type dependencies

data Action Source #

Actions that can be performed in FrozenLakeV0

Constructors

Left 
Down 
Right 
Up 

Instances

Bounded Action Source # 
Enum Action Source # 
Eq Action Source # 

Methods

(==) :: Action -> Action -> Bool #

(/=) :: Action -> Action -> Bool #

Ord Action Source # 
Show Action Source # 
Generic Action Source # 

Associated Types

type Rep Action :: * -> * #

Methods

from :: Action -> Rep Action x #

to :: Rep Action x -> Action #

Hashable Action Source # 

Methods

hashWithSalt :: Int -> Action -> Int #

hash :: Action -> Int #

ToJSON Action Source # 
(MonadThrow t, MonadIO t) => MonadEnv (EnvironmentT t) StateFL Action Reward Source # 
type Rep Action Source # 
type Rep Action = D1 (MetaData "Action" "Environments.Gym.ToyText.FrozenLakeV0" "reinforce-0.0.0.1-BYNakn0URySEY5wecxfdnO" False) ((:+:) ((:+:) (C1 (MetaCons "Left" PrefixI False) U1) (C1 (MetaCons "Down" PrefixI False) U1)) ((:+:) (C1 (MetaCons "Right" PrefixI False) U1) (C1 (MetaCons "Up" PrefixI False) U1)))