Returns the gradients on the Q-network for the different actions (used for policy update)

Tells whether the environment reached a terminal state after the last transition (i.e. the last transition that occured was terminal).

Prepare grid cells, walls. @param width grid's width. @param height grid's height. @param walls list of wall x,y tuples.

Getting the learning rate

Gets the most inner type (np.uint8, np.float32, ...) of [subject]. Parameters ----------- subject : int The subje

Gets a list of punctual observations composing this environment. This returns a list where element i is a punctual observation. Note

Called whenever the agent has chosen an action. This occurs after the agent state was updated with the new observation it made, but before it

Called whenever the agent has taken an action on its environment. This occurs after the agent applied this action on the environment and befo

Called whenever the agent ends an episode, just after this episode ended and before any onEpochEnd() signal could be sent. Parameters

Get the q values for one belief state Arguments --------- state_val : one belief state Returns -------

Get the average Q-values up to planning depth d for one pseudo-state. Arguments --------- state_val : array of objec

Get the q values for pseudo-state(s) with a planning depth d. This function is called recursively by decreasing the depth d at every step.

Returns a batch of states, actions, rewards, terminal status, and next_states for a number batch_size of randomly chosen transitions. Note tha

Return corresponding states, actions, rewards, terminal status, and next_states for a number batch_size of randomly chosen transitions. Note t

Resets the environment and put it in mode [mode]. This function is called when beginning every new episode. The [mode] can be used t

Reset environment for a new episode. Parameters ----------- Mode : int -1 corresponds to training and 0 to test

Reset environment for a new episode. Arguments: Mode : int -1 corresponds to training and 0 to test

Reset environment for a new episode. Arguments: Mode : int -1 corresponds to training and 0 to test

Get all rewards currently in the replay memory, ordered by time where they were received.

Activate or deactivate this controller. A controller should not react to any signal it receives as long as it is deactivated. For ins

Set all parameters used by the learning algorithm Arguments --------- list_of_values : list of numpy arrays lis

Solve maze, find path to ending cell. @returns path or None if not found.

Optional hook that can be used to show a summary of the performance of the agent on the environment in the current mode. Parameters

Train CRAR from one batch of data. Parameters ----------- states_val : numpy array of objects Each objec

Initializer. Parameters ----------- inputDims : list of tuples Each tuple relates to one of the observations where

The information contained in each node is: - Children and parent - Position: indice of the transition in the replay memory, i.e.

The tree does not implement any insert-related method because the idea is to initialize the tree to have the same number of leaves

Initializer.

Initializer.

Initializer.

Initializer.

Initializer.

Initializer.

Initialize the environment

Initialize environment

Initialize environment

Initialize environment. Arguments: rng - the numpy random number generator

Initialize environment Arguments: rng - the numpy random number generator

Initialize new cell. @param reachable is cell reachable? not a wall? @param x cell x coordinate @param y cell y coordinate

Initialize environment. Parameters ----------- rng : the numpy random number generator

Initialize environment. Parameters ----------- rng : numpy random number generator

Initialize environment. Arguments: rng - the numpy random number generator

Initialize environment. Arguments: rng - the numpy random number generator

Build a network consistent with each type of inputs

Perform one time step on the environment

Applies the agent action [action] on the environment. Parameters ----------- action : int The action selected by

Applies the agent action [action] on the environment. Parameters ----------- action : int The action selected by

Performs one time-step within the environment and updates the current observation self._last_ponctual_observation Parameters -------

Simulate one time step in the environment.

Simulate one time step in the environment.

Get the best action for a pseudo-state Arguments --------- state : list of numpy arrays One pseudo-state. The n

Get the best action for a pseudo-state Arguments --------- state : one pseudo-state Returns -------

Get the best action for a pseudo-state Arguments --------- state : one pseudo-state Returns -------